We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: Infrared Image Super-Resolution via Heterogeneous Convolutional WGAN

Abstract: Image super-resolution is important in many fields, such as surveillance and remote sensing. However, infrared (IR) images normally have low resolution since the optical equipment is relatively expensive. Recently, deep learning methods have dominated image super-resolution and achieved remarkable performance on visible images; however, IR images have received less attention. IR images have fewer patterns, and hence, it is difficult for deep neural networks (DNNs) to learn diverse features from IR images. In this paper, we present a framework that employs heterogeneous convolution and adversarial training, namely, heterogeneous kernel-based super-resolution Wasserstein GAN (HetSRWGAN), for IR image super-resolution. The HetSRWGAN algorithm is a lightweight GAN architecture that applies a plug-and-play heterogeneous kernel-based residual block. Moreover, a novel loss function that employs image gradients is adopted, which can be applied to an arbitrary model. The proposed HetSRWGAN achieves consistently better performance in both qualitative and quantitative evaluations. According to the experimental results, the whole training process is more stable.
Comments: To be published in the 18th Pacific Rim International Conference on Artificial Intelligence (PRICAI-2021)
Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Cite as: arXiv:2109.00960 [eess.IV]
  (or arXiv:2109.00960v1 [eess.IV] for this version)

Submission history

From: Yongsong Huang [view email]
[v1] Thu, 2 Sep 2021 14:01:05 GMT (17805kb,D)

Link back to: arXiv, form interface, contact.